OBJECTIVE: To assess reproducibility and validity of clinical classification of angina pectoris (AP) patients. DESIGN: Fifty-six patients scheduled for coronary angiography because of stable AP were classified by two independent observers with regard to (i) type and (ii) severity of chest pain (Canadian Cardiovascular Society, CCS) and (iii) cardiac functional status (New York Heart Association, NYHA). Myocardial perfusion imaging (MPI) was performed in 55 including measurement of ejection fraction in 46, angiography was undertaken in 51. RESULTS: Observers agreed 100% on the presence (n = 45) or absence (n = 11) of angina. They agreed in 52 (93%), 48 (86%), and 42 (75%) patients with regard to type of pain, CCS grade, and NYHA class, respectively. In the remaining patients, they disagreed by one class only. The positive and negative predictive values of typical/atypical angina for perfusion abnormalities and coronary disease were 55%/82% and 53%/ 82%, respectively. CONCLUSIONS: Observer agreement was excellent for presence, type, and severity of chest pain but moderate with regard NYHA class. Clinical judgment could not predict with reasonable accuracy abnormal perfusion or coronary artery disease.